A high fidelity simulator for a quadrotor UAV using ROS and Gazebo

Abstract

Flight tests of prototype UAV systems can be restricted by spatial constraints and they may bring risks of damage due to failures. Motivated by these, we presented a simulation approach based on Robot Operating System (ROS) and Gazebo. Unlike other state-of-the-art quadrotor simulators, we implemented the dynamics model of the UAV in ROS to achieve high fidelity behavior of the UAV. A hierarchical navigation system is also presented in our paper. The system layers include simultaneous localization and mapping (SLAM), mapping framework in Cartesian and polar coordinates, A* global path planner, revised vector field histogram plus (VFH+) for optimal local path selection and online trajectory algorithm (OTA) with collision checking for obstacle avoidance. In order to cater for vision-based applications, quadrotor is equipped with a monocular camera in the simulation model. The implementation of circle and landing pad detection and tracking algorithm demonstrates the functionality of vision guidance. In our simulation, various aspects including complex indoor and outdoor environments and on-board sensors are capable of simultaneously interacting with our navigation system to achieve certain surveillance missions. In the end, we demonstrated the applicability of our complex quadrotor systems by performing an autonomous navigation task in simulated complex environments. In comparison with the experimental data, simulation results align with the ones in flight tests in terms of real flight behaviors during navigation tasks in general.